ZnO-TiO2 nanocomposite core-shell

7
ZnOTiO 2 Nanocomposite Films for High Light Harvesting Eciency and Fast Electron Transport in Dye-Sensitized Solar Cells Venkata Manthina, ,§ Juan Pablo Correa Baena, ,§ Guangliang Liu, ,§ and Alexander G. Agrios* ,,§ Department of Civil & Environmental Engineering, University of Connecticut, Unit 3037, 261 Glenbrook Rd, Storrs, Connecticut 06269, United States Department of Chemical, Materials & Biomolecular Engineering, University of Connecticut, Unit 3222, 191 Auditorium Rd, Storrs, Connecticut 06269, United States § Center for Clean Energy Engineering, University of Connecticut, 44 Weaver Rd, Storrs, Connecticut 06269, United States ABSTRACT: Electron transport and recombination are the essential processes that determine the charge collection eciency in dye-sensitized solar cells (DSSC). While nearly 100% of charges are collected in well-built ordinary DSSCs, this value can be sharply reduced by the use of redox couples other than iodide/triiodide due to fast electron recombination. To compensate, structures capable of fast electron transport are needed. Nanorod arrays that have this attribute tend to suer from low surface area, resulting in low dye loading and reduced light harvesting. We have therefore developed a novel nanocomposite structure consisting of zinc oxide (ZnO) nanorods coated with titanium dioxide (TiO 2 ) nanoparticles using an electrostatic layer-by-layer (LbL) deposition technique. The titanium dioxide nanoparticle coating can add an order of magnitude of surface area and is compatible with known high- performance dyes. This composite nanostructure has been designed to take advantage of the improved electron transport along the nanorods and surface area provided by the nanoparticles, yielding good charge collection and light harvesting. Transient measurements indicate that the composite lm can transport electrons at least 100 times faster than a nanoparticulate TiO 2 lm. In tests using ferrocene/ferrocenium as a model alternative redox couple with fast recombination, currentvoltage measurements indicate that the ZnOTiO 2 hybrid lms generate much higher currents than conventional TiO 2 nanoparticulate lms. However, not all charges successfully transfer from TiO 2 to ZnO due to an energy barrier between the materials. INTRODUCTION Dye-sensitized solar cells (DSSCs) 1 are low-cost alternatives to silicon photovoltaics. The conventional DSSC consists of two sandwiched pieces of conducting glass, one of them coated with a mesoporous layer of nanoparticulate TiO 2 with a self- assembled monolayer of chemisorbed dye molecules, lled with an electrolyte for dye regeneration. The dye is a transition- metal complex or organic chromophore that harvests sunlight by absorbing strongly in the visible region of the solar spectrum. The principal photovoltaic losses in the DSSC are due to incomplete light harvesting, recombination of the photoinjected electrons with the electrolyte, and the over- potential required for dye regeneration. Ruthenium complex dyes like N719, N3, and black dyeexhibit high eciency with the I /I 3 redox couple, but there is little room for improvement in light harvesting in the visible range. In DSSCs, the I /I 3 redox couple limits the overall eciency due to the high overpotential (ca. 0.5 V) for dye regeneration by I . This is believed to be due to the complex multielectron mechanism of the I /I 3 redox couple involving the radicals diiodide (I 2 ) and atomic iodine (I ). 2 The eect is that the dye HOMO level must be about 0.5 V more positive than the I /I 3 redox level. Since the voltage output of the cell is the dierence between the redox level and the quasi-Fermi level in the semiconductor, this 0.5 V is lost. An alternative redox electrolyte that could regenerate the sensitizer from a potential closer to its HOMO would result in higher cell voltage and eciency, if all else were held equal. Unfortunately, alternatives (such as iron or cobalt complexes) tend to recombine rapidly with conduction-band electrons, reducing the quasi-Fermi level and resulting in no benet. There have been important recent advances in using specialized dyes to slow recombination with electrolytes based on ferrocene or cobalt bipyridyl complexes. 35 Our approach is to utilize novel semiconductor structures to attain fast electron transport in order to improve charge collection eciency. The combination of these approaches will maximize solar cell performance. In particular, by providing some tolerance of accelerated recombination, structures for fast electron transport enable the use of standard dyes such as N719, which are more facile products than the specialized dyes Received: May 12, 2012 Revised: September 22, 2012 Published: October 2, 2012 Article pubs.acs.org/JPCC © 2012 American Chemical Society 23864 dx.doi.org/10.1021/jp304622d | J. Phys. Chem. C 2012, 116, 2386423870

Transcript of ZnO-TiO2 nanocomposite core-shell

ZnO−TiO2 Nanocomposite Films for High Light Harvesting Efficiencyand Fast Electron Transport in Dye-Sensitized Solar CellsVenkata Manthina,†,§ Juan Pablo Correa Baena,†,§ Guangliang Liu,‡,§ and Alexander G. Agrios*,†,§

†Department of Civil & Environmental Engineering, University of Connecticut, Unit 3037, 261 Glenbrook Rd, Storrs, Connecticut06269, United States‡Department of Chemical, Materials & Biomolecular Engineering, University of Connecticut, Unit 3222, 191 Auditorium Rd, Storrs,Connecticut 06269, United States§Center for Clean Energy Engineering, University of Connecticut, 44 Weaver Rd, Storrs, Connecticut 06269, United States

ABSTRACT: Electron transport and recombination are theessential processes that determine the charge collectionefficiency in dye-sensitized solar cells (DSSC). While nearly100% of charges are collected in well-built ordinary DSSCs,this value can be sharply reduced by the use of redox couplesother than iodide/triiodide due to fast electron recombination.To compensate, structures capable of fast electron transportare needed. Nanorod arrays that have this attribute tend tosuffer from low surface area, resulting in low dye loading andreduced light harvesting. We have therefore developed a novelnanocomposite structure consisting of zinc oxide (ZnO)nanorods coated with titanium dioxide (TiO2) nanoparticles using an electrostatic layer-by-layer (LbL) deposition technique.The titanium dioxide nanoparticle coating can add an order of magnitude of surface area and is compatible with known high-performance dyes. This composite nanostructure has been designed to take advantage of the improved electron transport alongthe nanorods and surface area provided by the nanoparticles, yielding good charge collection and light harvesting. Transientmeasurements indicate that the composite film can transport electrons at least 100 times faster than a nanoparticulate TiO2 film.In tests using ferrocene/ferrocenium as a model alternative redox couple with fast recombination, current−voltage measurementsindicate that the ZnO−TiO2 hybrid films generate much higher currents than conventional TiO2 nanoparticulate films. However,not all charges successfully transfer from TiO2 to ZnO due to an energy barrier between the materials.

■ INTRODUCTION

Dye-sensitized solar cells (DSSCs)1 are low-cost alternatives tosilicon photovoltaics. The conventional DSSC consists of twosandwiched pieces of conducting glass, one of them coated witha mesoporous layer of nanoparticulate TiO2 with a self-assembled monolayer of chemisorbed dye molecules, filled withan electrolyte for dye regeneration. The dye is a transition-metal complex or organic chromophore that harvests sunlightby absorbing strongly in the visible region of the solarspectrum. The principal photovoltaic losses in the DSSC aredue to incomplete light harvesting, recombination of thephotoinjected electrons with the electrolyte, and the over-potential required for dye regeneration. Ruthenium complexdyes like N719, N3, and “black dye” exhibit high efficiency withthe I−/I3

− redox couple, but there is little room forimprovement in light harvesting in the visible range.In DSSCs, the I−/I3

− redox couple limits the overallefficiency due to the high overpotential (ca. 0.5 V) for dyeregeneration by I−. This is believed to be due to the complexmultielectron mechanism of the I−/I3

− redox couple involvingthe radicals diiodide (I2

•−) and atomic iodine (I•).2 The effectis that the dye HOMO level must be about 0.5 V more positivethan the I−/I3

− redox level. Since the voltage output of the cell

is the difference between the redox level and the quasi-Fermilevel in the semiconductor, this 0.5 V is lost. An alternativeredox electrolyte that could regenerate the sensitizer from apotential closer to its HOMO would result in higher cellvoltage and efficiency, if all else were held equal. Unfortunately,alternatives (such as iron or cobalt complexes) tend torecombine rapidly with conduction-band electrons, reducingthe quasi-Fermi level and resulting in no benefit.There have been important recent advances in using

specialized dyes to slow recombination with electrolytesbased on ferrocene or cobalt bipyridyl complexes.3−5 Ourapproach is to utilize novel semiconductor structures to attainfast electron transport in order to improve charge collectionefficiency. The combination of these approaches will maximizesolar cell performance. In particular, by providing sometolerance of accelerated recombination, structures for fastelectron transport enable the use of standard dyes such asN719, which are more facile products than the specialized dyes

Received: May 12, 2012Revised: September 22, 2012Published: October 2, 2012

Article

pubs.acs.org/JPCC

© 2012 American Chemical Society 23864 dx.doi.org/10.1021/jp304622d | J. Phys. Chem. C 2012, 116, 23864−23870

mentioned above, which tend to be large and difficult tosynthesize.Electron transport in DSSCs is slow, due to trapping of the

electrons in the grain boundaries and the relatively long andtortuous path of the electron to the fluorine-doped tin oxide(FTO).6,7 In a 10 μm thick film an electron visits about 106

nanoparticles on average before reaching the FTO surface.8

The resulting slow transport is adequate in the presence ofiodide/triiodide, since its recombination kinetics are slow, butinadequate when using an alternative redox couple with fasterrecombination.Fast electron transport can be achieved by developing 1-D

nanostructures like nanotubes, nanorods, and nanowires ofmetal oxides. ZnO is a highly favorable material for applicationin DSSCs, since it can be grown in monocrystalline nanorodsusing facile methods, its electron mobility is high,9 and its bandedge energies are very close to those of TiO2.

10 ZnO is themetal oxide with the second highest efficiency achieved inDSSCs.11−13 ZnO nanorods can be synthesized on varioussubstrates in situ by procedures such as chemical bathdeposition,14,15 electrodeposition,16−18 and chemical vapordeposition,19,20 and in DSSCs they can transport electronsmore than an order of magnitude faster than a TiO2nanoparticulate film.21 However, the main disadvantage ofnanorods is their lower surface area than a nanoparticle film foradsorption of light-harvesting molecules. In addition, dyespartially dissolve the ZnO, forming a deleterious Zn2+−dyesurface complex.22

Previous efforts to circumvent these shortcomings havetended to either boost the surface area of the ZnO, e.g. byadding ZnO nanoparticles23 or secondary nanorods,24 or toimprove dye performance via, for example, a protectiveconformal TiO2 encapsulation of ZnO.25 To create a structurewith both fast electron transport and high surface area, wedeveloped a hybrid photoanode consisting of ZnO nanorodscoated with TiO2 nanoparticles using facile wet-chemicalmethods. ZnO nanorods were grown by chemical bathdeposition (CBD), a low-temperature catalyst-free processsuitable for industrial applications. TiO2 nanoparticles werecoated uniformly over the nanorods by electrostatic layer-by-layer (LbL) deposition. The LbL technique is based onexposing a substrate sequentially to cationic and anionicsubstances, which form successive bilayers by electrostaticattraction. The TiO2 nanoparticles greatly increase the totalsurface area and provide a superior surface for dye attachmentcompared to ZnO, while the nanorods can provide fasttransport of electrons from TiO2 to the conducting substrate.In addition, the structure includes wide open channels that arehelpful for mass transport of redox species and for filling withsolid hole-transporting materials in solid-state DSSC devices.Despite the simplicity and attractiveness of this structure andfabrication method, it has not been done previously, althoughWang et al. added a thin layer of 5-nm TiO2 particles to ZnOnanorods by sputtering.26

In this work, we compared hybrid films to ZnO nanorod-only films and TiO2 nanoparticle-only films in terms of dyeloading and device performance. For the latter measure wecompared DSSC devices using two different redox couples:iodide/triiodide (I−/I3

−) and ferrocene/ferrocenium (Fc/Fc+).We have chosen the latter as a model alternative redox couplewith fast recombination kinetics. Our tests with Fc/Fc+ showthat these hybrid photoanodes can collect more injected

electrons than a conventional TiO2 nanoparticle film withequivalent dye loading.

■ EXPERIMENTAL METHODSReagents and Materials. Except where noted, all

chemicals were purchased from Sigma-Aldrich and were ACSgrade or better. N719 was purchased from Dyesol. SnO2:F glass(FTO, transmission >80% in the visible spectrum; sheetresistance 8 Ω/□) was purchased from Hartford Glass Co.

Electrode Fabrication. ZnO nanorods were synthesized bya two-step chemical bath deposition (CBD) technique.27 Thenanorods were optimized for subsequent TiO2 deposition usinga modified seed layer technique that will be the subject of afuture publication. Slides of borosilicate glass with a conductingSnO2:F (FTO) layer were cut into 25 × 50 mm pieces andcleaned by 10 min sonication in a detergent solution (5% RBS-25 in water) followed by ethanol. The clean FTO substrateswere coated with the seed layer and then heated at 350 °C on atitanium hot plate for 30 min. CBD was followed by immersionin an aqueous solution of 50 mM zinc nitrate hexahydrate, 50mM hexamethylenetetramine, and 6 mM polyethyleneimine.The seeded substrates were placed at an angle of 60° fromhorizontal in a 100-mL glass bottle with the seeds facing thebottom of the bottle and held in an oven at 90 °C for 24 h.TiO2 nanoparticles were synthesized by hydrolysis of

titanium tetraisopropoxide as previously described.28,29 Electro-static layer-by-layer (LbL) deposition was accomplished byimmersing a substrate alternately in a cationic 1 g/L polymersolution (PDAC, polydiallyldimethyl ammonium chloride, 70kDa) in water and in an aqueous suspension of anionic TiO2particles with intermediate rinsing and drying steps, using adipping robot (DR-3, Riegler & Kirstein GmbH).29 Allsolutions contained 5 mM triethylamine (resulting in a pHnear 11) to maintain negative particle surface charge. Fordeposition on bare FTO, the dipping time in the TiO2suspension was 5 min. When coating nanorods, the dippingtime was 30 min to allow penetration of the channels betweenthe rods. LbL was followed by sintering at 500 °C to removepolymer layers.

Surface/Structure Characterization. The morphology ofZnO nanorods and TiO2 nanoparticles was investigated byscanning electron microscopy (FEI Quanta FEG250 SEM inhigh vacuum mode). The ZnO nanorods and TiO2 nano-particles were additionally characterized by X-ray powderdiffraction (XRD) using a Bruker D8 Advance X-raydiffractometer using Cu Kα radiation (λ = 0.154 178 nm) ata scanning rate of 0.04° s−1 in the 2θ range from 10° to 90°.The XRD data were analyzed using the Debye−Scherrerequation to determine the TiO2 crystallite size.

Sensitization. After sintering, films were allowed to cool to100 °C and then immediately immersed in 0.3 mM N719 inethanol. After 12 h they were removed, rinsed in acetonitrile,and dried in air. To measure the amount of N719 dye adsorbedby a film, a dyed and twice-rinsed 4.5 cm ×2.5 cm electrode wasplaced in 3 mL of 0.1 M NaOH in water. As the solvent volumewas not sufficient to immerse the electrode, solvent wasrepeatedly pipetted over the electrode. After repetitions, thedye was visibly completely desorbed from the electrode. Thespectrum of the desorbed dye was measured in a 1-cm quartzcuvette using a Varian Cary 50 spectrophotometer. Theliterature peak extinction coefficient of N719 (1.43 × 104

M−1 cm−1) in ethanol30 was used to obtain the correctedvalue in basified water. The molar extinction coefficient for

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp304622d | J. Phys. Chem. C 2012, 116, 23864−2387023865

N719 in water with 0.1 M NaOH was found to be 1.65 × 104

M−1 cm−1. This value was used to quantify the amount ofdesorbed dye.Solar Cell Assembly. Each sensitized electrode was sealed

against a counter electrode on a hot plate at 120 °C using a hot-melt plastic frame (Solaronix, Meltonix 1170, 25 μm thick), byapplying light pressure with a glass rod. The assembled cell wasfilled with electrolyte through two holes in the counterelectrode. The holes were then sealed using hot-melt plasticand a thin glass cover slide. The exposed conducting glass leadsof each electrode were coated with copper tape (3M) forimproved electrical conductivity.Electrolyte Composition. Minimal electrolyte recipes

were used to exclude complications due to interactions withthe various additives that are commonly used. Iodide/triiodide(I−/I3

−) electrolyte was prepared with 0.5 M tetrabutylammo-nium iodide and 0.05 M iodine (I2) in 3-methoxypropionitrile.The ferrocene/ferrocenium (Fc/Fc+) electrolyte contained 0.1M ferrocene and 0.05 M ferrocenium hexafluorophosphate(Aldrich) in 3-methoxypropionitrile. The Fc/Fc+ electrolytewas prepared fresh and deoxygenated by bubbling nitrogen 10min prior to cell fabrication to minimize reaction of ferrocenewith oxygen.31

Transient Measurements. Measurements of electrontransport time were made using a set of National Instrumentscomponents in a PXIe chassis capable of high-resolution analogvoltage output and digitized input. A square-wave modulationwas applied to a white-light LED that was used to illuminate aDSSC. The modulation amplitude produced a <10% change inDSSC current, which is linear with light intensity. The DSSCwas operated in series with a 65-Ω resistor and the voltageacross this resistor was measured. The current is determined byI = V/R. We refer to this as a “quasi-short-circuit” mode, as thesmall voltages measured (<20 mV) were close to the short-circuit condition, compared to open circuit voltages ofhundreds of millivolts. At least 50 transients were averagedfor noise reduction.Solar Cell Characterization. Current−voltage (J−V)

measurements were made using a Keithley 2400 source/meter controlled by a PC, while irradiating at 100 mW/cm2 (1sun) with AM 1.5G simulated sunlight produced by a solarsimulator (Newport 91160), calibrated against a siliconreference cell with KG5 filter (PV Measurements, Inc., Boulder,CO). The DSSC active area was 1 cm2.

■ RESULTS AND DISCUSSIONZnO Nanorods. ZnO nanorods were fabricated on FTO. A

typical example is shown in Figure 1, having nanoroddimensions of about 5 μm length and 600 nm diameter anda number density of 1.2 × 108 rods per cm2 of FTO after 24 hdeposition. The XRD data (Figure 2a) show that the nanorodsgrew as crystalline ZnO with the hexagonal wurtzite structure[space group P63mc (186); a = 0.3249 nm, c = 0.5206 nm].The data are in agreement with the Joint Committee onPowder Diffraction Standards (JCPDS) card for ZnO (JCPDS070-8070). The prominence of the peak assigned to the (002)plane of ZnO is consistent with predominant ZnO growthalong the c-axis, perpendicular to the substrate, giving thenanorod morphology.ZnO−TiO2 Hybrid Films. TiO2 nanoparticles were

synthesized by an autoclave hydrolysis method and ultimatelydispersed in water. On the basis of a Scherrer analysis of the(010) peak at 26° of an X-ray powder diffraction pattern (not

Figure 1. SEM images of ZnO nanorods in top view (a) and crosssection (b). Scale bar: 1 μm.

Figure 2. X-ray diffraction pattern of (a) ZnO nanorods and (b) ZnOnanorods coated with TiO2 nanoparticles, on a fluorine-doped tinoxide substrate. Z and A denote wurtzite ZnO and anatase TiO2,respectively.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp304622d | J. Phys. Chem. C 2012, 116, 23864−2387023866

shown), the crystallite size of the TiO2 nanoparticles was foundto be 16 nm, and only the anatase phase was observed. Particlesize is generally similar to crystallite size for this synthesismethod.32 LbL deposition was used to deposit these nano-particles over ZnO nanorods. Initial trials used a TiO2deposition time of 5 min per LbL cycle. XRD spectra forthese samples (Figure 2b) have added peaks indicating theanatase phase of TiO2, while retaining the same ZnO wurtzitepeaks, indicating that the ZnO nanorod crystal structures werenot affected by the deposition and sintering.Scanning electron micrographs showing the results of

different numbers of LbL layers of TiO2 are shown in Figure3. The first layer of TiO2 coats the nanorods well, with

successive layers increasing the TiO2 film thickness while aconformal coating is maintained. However, cross-sectionalimages (not shown) revealed that the TiO2 particles did notreach the bottom of the nanorods. We therefore lengthened theTiO2 dipping time to 30 min in each LbL cycle. With thismodification, the TiO2 nanoparticles completely penetrate thenanorod film and evenly coat all surfaces. An example usingthree layers is shown in top view and in cross section, in Figure4.Dye Loading. If small spheres form a monolayer with

simple square packing over a large cylinder of area A (notconsidering the end faces), the added area will be πA. Thisimplies that three monolayers would increase the surface areaby about an order of magnitude. Dye desorption measurements(Figure 5) following sensitization with N719 showed that aZnO nanorod array (Z) adsorbed 3.5 nmol of dye/cm2 ofgeometric film area, whereas a nanorod array coated with threelayers of TiO2 nanoparticles (ZT) adsorbed 13.9 nmol/cm2, fora difference of nearly a factor of 4. That this is less than thefactor of ∼10 calculated above is not surprising given theapproximation of square packing and the fact that an LbL filmhas considerable porosity.29 We found that 30 layers of TiO2deposited on a flat FTO substrate had a dye loadingcomparable to that of the ZT hybrid film. Therefore, 30-layerTiO2 films (T) were used in the device performance

experiments for comparison with the composite films. Thethickness of the 30-layer TiO2 film is approximately 0.6 μm asmeasured by scanning electron microscopy (not shown). LbLTiO2 films have similar morphology to doctor-bladed TiO2films29 and similar performance in DSSCs when deposited atcomparable thickness.33 For reference, a dye loading for atypical 10-μm film would be ca. 100 nmol/cm2.34,35

Electron Transport. Figure 6a shows representative currenttransients at quasi-short-circuit conditions for Z, ZT, and Tfilms. Each transient is well-fit by a single exponential decay of

Figure 3. SEM images of ZnO nanorods coated with zero (a), one (b),three (c), or five (d) layers of TiO2 with 5-min immersion periods.Scale bar: 1 μm.

Figure 4. ZnO nanorods coated with three layers of TiO2nanoparticles using 30-min dipping times shown in top view (a) andcross section (b). Scale bar: 1 μm.

Figure 5. Uptake of N719 by a ZnO nanorod array (Z), a nanorodarray coated with three layers of TiO2 nanoparticles (ZT), and film of30 layers of TiO2 nanoparticles (T), all on FTO.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp304622d | J. Phys. Chem. C 2012, 116, 23864−2387023867

the form y = y0 + A exp(−t/τtr), where τtr is the characteristictime for electron transport. The fitted time constants τtr over arange of light intensities are plotted against the correspondingshort-circuit current density JSC in Figure 6b. For TiO2, thepower-law decrease of τtr with increasing JSC is typical of DSSCsand is well-described by a trapping−detrapping model.36 Timeconstants for the other samples are relatively constant at about0.29 ms for ZnO nanorods and 0.46 ms for ZnO nanorodscoated with TiO2 nanoparticles. The invariability of τtr with JSCfor the Z and ZT samples strongly suggests that a limit of themeasurement has been reached, most likely due to an RC timebased on the resistance of the cell and resistor connected inseries and the capacitance of the semiconductor/electrolyteinterface.21 These time constants should therefore be taken asan upper limit on the electron transport time of the ZnOnanorod-based materials. Since the RC time may differ fordifferent samples, the results shown in Figure 6b cannotdifferentiate between the τtr value of Z and ZT films, but theydo demonstrate that electron transport in these samples is atleast 2 orders of magnitude faster than transport throughnanoparticulate TiO2.Device Performance. Figure 7a compares the current−

voltage characteristics of the DSSCs made from the threedifferent films in a conventional iodide/triiodide electrolyte.Each trace represents the mean of triplicate measurements, andthe dotted lines indicate one standard deviation from the meanat each voltage. The ZT hybrid film strongly outperforms the Zfilm, indicating that the higher surface area and dye loading ofthe ZT film translate into higher photocurrents. Under theseconditions, the TiO2-only film T outperforms both Z and ZT,

which is to be expected since TiO2 nanoparticles are known togive the best performance of any material when recombinationrates are low.The benefit of the hybrid structure is revealed when using

ferrocene/ferrocenium as a model alternative redox couple withhigh recombination rates (Figure 7b). We focus here on thechanges in the short circuit current (JSC), as these report on theelectron collection efficiency. We note that faster electrontransport can improve JSC due to higher electron collectionefficiency, but it cannot improve the open-circuit voltage (VOC)since no electron transport occurs at open circuit. Values of JSCextracted from Figure 7 are compared in Figure 8. The rapid

Figure 6. (a) Transients for decay of quasi-short-circuit current inresponse to square-wave modulation of light intensity, with single-exponential fits. The inset shows TiO2 transient at a longer time scale.(b) Fitted electron transport time constants versus quasi-short-circuitcurrent for T, ZT, and Z films.

Figure 7. Current−voltage (J−V) characteristics for solar cells usingZT, Z, and T films and electrolyte based on (a) I−/I3

− or (b) Fc/Fc+,under illumination with 100 mW/cm2 AM1.5G simulated sunlight.Solid and dotted lines show mean ± SD for triplicate samples. Theinset in part b magnifies the T trace.

Figure 8. Short circuit current of DSSCs constructed with I−/I3− and

Fc/Fc+ electrolyte using Z, ZT (hybrid), and T films underillumination with 100 mW/cm2 AM1.5G simulated sunlight.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp304622d | J. Phys. Chem. C 2012, 116, 23864−2387023868

scavenging of conduction band electrons by Fc+ almostcompletely short-circuits a cell with a conventional TiO2nanoparticulate film (T), resulting in a barely visible J−Vtrace with a short-circuit current of only 16 μA/cm2. A ZnOnanorod film (Z) has much better performance, with JSC = 0.72mA/cm2. The hybrid film ZT has the best short circuit current(0.93 mA/cm2) due to its higher surface area and dye loading.The difference in performance here is smaller than with I−/I3

−,which we attribute to (1) recombination of electrons from theTiO2 with the Fc+ during the time between electron injectionand electron transfer from TiO2 to the ZnO, and/or (2)impeded electron transfer from the TiO2 into the ZnO, as willbe discussed below. The surprising result that the Z film obtainsa higher current in an Fc/Fc+ electrolyte than in an I−/I3

electrolyte may be the result of corrosion of ZnO by the latterelectrolyte,37 which we have observed in SEM images (notshown) of ZnO nanorods before and after exposure to I−/I3

−.Energy Barrier. Transfer of electrons from the TiO2

nanoparticles to the ZnO nanorods must be downhill inenergy. Assuming that electron transfer occurs between theconduction bands of the two materials and that electrons arerapidly thermalized within the conduction bands, this requiresthat the conduction band minimum (CBM) of ZnO lie at amore positive potential than that of TiO2. The conduction bandedges of ZnO and TiO2 would be expected to lie quite close toeach other in energy, and the relative difference between thetwo is difficult to predict. As an indirect measure of this bandedge offset, we measured current−voltage curves in the darkand in the absence of dye. The measured “dark current” is dueentirely to recombination at the semiconductor/electrolyteinterface.In this measurement, electrons flow from the FTO into the

semiconductor and reduce triiodide to iodide, which is oppositein sense to a photocurrent at voltages below VOC. As TiO2nanoparticles are added to ZnO nanorods, if the CBM of ZnOis more negative than that of TiO2, then electrons shouldtransfer easily from ZnO into TiO2, and the added surface areaprovided by the TiO2 nanoparticles should increase the darkcurrent. On the other hand, if TiO2 has the more negativeCBM, electron transfer into TiO2 should be impeded, and theTiO2 should reduce the dark current by reducing the ZnOsurface area that is in contact with the electrolyte. The resultshown in Figure 9a is that the ZT film has more dark currentthan the Z film, which is consistent with ZnO having the morenegative CBM. The dark current from ZT is close to that of aTiO2-only film (T) of comparable surface area, based on theabove dye desorption measurements. We note that some of thedifferences seen in Figure 9 may reflect different reactivities ofZnO and TiO2 toward I3

−, as suggested by the longer electronlifetime observed in ZnO.9

For the reverse experiment, we grew short (1-μm) ZnOnanorods on a TiO2 film by adding two layers of a standardseed layer (drop-coating 5 mM zinc acetate in ethanol, allowingto evaporate, and heating at 350 °C) to a TiO2 film followed byimmersion in a ZnO CBD solution (the same as for the longerrods) at 90 °C for 4 h (Figure 10). This TZ hybrid film shouldhave more dark current than a plain T film if and only if ZnOhas the more positive CBM. On the contrary, the addition ofZnO significantly reduces the dark current (Figure 9b). Bothsets of results, then, are consistent with a more negative CBMfor ZnO, which would tend to pose a barrier to the desiredelectron transfer from TiO2 to ZnO (Figure 11). Given thegood TiO2 coverage in the ZT films (see Figure 4), it is likely

that electrons photoinjected into TiO2 can travel to theconducting glass substrate along a purely TiO2 path on theoutside of the nanorod. This allows high current in ZT samplesin an I−/I3

− electrolyte, where electron diffusion lengths38 aremuch longer than the nanorods in these samples (about 5 μm).However, in the presence of Fc/Fc+, recombination claimsmany of these electrons, reducing the benefit of the TiO2nanoparticulate coating. The fact that ZT still outperforms Z(and T) in a Fc/Fc+ electrolyte suggests that some electrons areable to transfer from TiO2 to ZnO, perhaps due to a smallenergy barrier that can be thermally overcome by a fraction ofthe conduction band electrons at room temperature, resultingin slow transfer kinetics. Our laboratory is currently at work onstrategies to eliminate this energy barrier.

Figure 9. Dark currents of dye-free films: (a) Z, ZT, and T and (b) TZand T.

Figure 10. ZnO nanorods (1 μm) grown atop a 30-layer LbL TiO2nanoparticulate film. Scale bar: 1 μm.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp304622d | J. Phys. Chem. C 2012, 116, 23864−2387023869

■ CONCLUSIONWe have successfully synthesized a composite nanostructureconsisting of ZnO nanorods coated with TiO2 nanoparticles,using facile wet-chemical methods. A coating of threemonolayers of TiO2 nanoparticles increased the dye uptakeof the film by a factor of 4 compared to ZnO nanorods only.Although a TiO2-only film of similar dye loading outperformsthe ZnO−TiO2 film with I−/I3

− electrolyte, the nanocompositefilm dramatically outperforms the TiO2 film when using thefast-recombining Fc/Fc+ electrolyte and modestly outperformsthe ZnO nanorod-only film. Transfer of electrons from theTiO2 to the ZnO is impeded by an energy barrier, as theconduction band minimum of ZnO lies at a slightly morenegative potential than that of TiO2, and more work is neededto remove this barrier. This nanocomposite photoanode pointsto a class of structures that can provide both fast transport andhigh surface area, enabling DSSCs that are tolerant ofelectrolytes with higher recombination rates than are foundwith the conventional iodide/triiodide couple.

■ AUTHOR INFORMATIONCorresponding Author*E-mail: [email protected] authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was funded by the University of ConnecticutResearch Foundation. The authors thank Dr. Teresa Lana-Villarreal for providing preliminary samples of ZnO nanorods.

■ REFERENCES(1) O’Regan, B.; Gratzel, M. Nature 1991, 353, 737.(2) Boschloo, G.; Hagfeldt, A. Acc. Chem. Res. 2009, 42, 1819.(3) Feldt, S. M.; Gibson, E. A.; Gabrielsson, E.; Sun, L.; Boschloo, G.;Hagfeldt, A. J. Am. Chem. Soc. 2010, 132, 16714.(4) Daeneke, T.; Kwon, T.-H.; Holmes, A. B.; Duffy, N. W.; Bach,U.; Spiccia, L. Nat. Chem. 2011, 3, 211.(5) Yella, A.; Lee, H.-W.; Tsao, H. N.; Yi, C.; Chandiran, A. K.;Nazeeruddin, M. K.; Diau, E. W.-G.; Yeh, C.-Y.; Zakeeruddin, S. M.;Gratzel, M. Science 2011, 334, 629.(6) Bisquert, J.; Cahen, D.; Hodes, G.; Ruhle, S.; Zaban, A. J. Phys.Chem. B 2004, 108, 8106.(7) Nissfolk, J.; Fredin, K.; Hagfeldt, A.; Boschloo, G. J. Phys. Chem. B2006, 110, 17715.

(8) Kopidakis, N.; Benkstein, K. D.; van de Lagemaat, J.; Frank, A. J.J. Phys. Chem. B 2003, 107, 11307.(9) Quintana, M.; Edvinsson, T.; Hagfeldt, A.; Boschloo, G. J. Phys.Chem. C 2007, 111, 1035.(10) Keis, K.; Bauer, C.; Boschloo, G.; Hagfeldt, A.; Westermark, K.;Rensmo, H.; Siegbahn, H. J. Photochem. Photobiol., A 2002, 148, 57.(11) Xu, F.; Sun, L. Energy Environ. Sci. 2011, 4, 818.(12) Keis, K.; Magnusson, E.; Lindstrom, H.; Lindquist, S.-E.;Hagfeldt, A. Sol. Energy Mater. Sol. Cells 2002, 73, 51.(13) Saito, M.; Fujihara, S. Energy Environ. Sci. 2008, 1, 280.(14) O’Brien, P.; Saeed, T.; Knowles, J. J. Mater. Chem. 1996, 6, 1135.(15) Vayssieres, L.; Keis, K.; Hagfeldt, A.; Lindquist, S.-E. Chem.Mater. 2001, 13, 4395.(16) Yoshida, T.; Tochimoto, M.; Schlettwein, D.; Wohrle, D.;Sugiura, T.; Minoura, H. Chem. Mater. 1999, 11, 2657.(17) O’Regan, B.; Sklover, V.; Gratzel, M. J. Electrochem. Soc. 2001,148, C498.(18) Xu, L.; Chen, Q.; Xu, D. J. Phys. Chem. C 2007, 111, 11560.(19) Bacsa, R. R.; Dexpert-Ghys, J.; Verelst, M.; Falqui, A.; Machado,B.; Bacsa, W. S.; Chen, P.; Zakeeruddin, S. M.; Graetzel, M.; Serp, P.Adv. Funct. Mater. 2009, 19, 875.(20) Huang, M. H.; Wu, Y.; Feick, H.; Tran, N.; Weber, E.; Yang, P.Adv. Mater. 2001, 13, 113.(21) Galoppini, E.; Rochford, J.; Chen, H.; Saraf, G.; Lu, Y.; Hagfeldt,A.; Boschloo, G. J. Phys. Chem. B 2006, 110, 16159.(22) Horiuchi, H.; Katoh, R.; Hara, K.; Yanagida, M.; Murata, S.;Arakawa, H.; Tachiya, M. J. Phys. Chem. B 2003, 107, 2570.(23) Ku, C.-H.; Wu, J.-J. Nanotechnology 2007, 18, 505706.(24) Park, J. Y.; Choi, S.-W.; Asokan, K.; Kim, S. S. J. Am. Ceram. Soc.2010, 93, 3190.(25) Law, M.; Greene, L. E.; Radenovic, A.; Kuykendall, T.; Liphardt,J.; Yang, P. J. Phys. Chem. B 2006, 110, 22652.(26) Wang, M.; Huang, C.; Cao, Y.; Yu, Q.; Deng, Z.; Liu, Y.; Huang,Z.; Huang, J.; Huang, Q.; Guo, W.; Liang, J. J. Phys. D: Appl. Phys.2009, 42, 155104.(27) Vayssieres, L. Adv. Mater. 2003, 15, 464.(28) Wang, P.; Zakeeruddin, S. M.; Comte, P.; Charvet, R.;Humphry-Baker, R.; Gratzel, M. J. Phys. Chem. B 2003, 107, 14336.(29) Agrios, A. G.; Cesar, I.; Comte, P.; Nazeeruddin, M. K.; Gratzel,M. Chem. Mater. 2006, 18, 5395.(30) Nazeeruddin, M. K.; Zakeeruddin, S. M.; Humphry-Baker, R.;Jirousek, M.; Liska, P.; Vlachopoulos, N.; Shklover, V.; Fischer, C.-H.;Gratzel, M. Inorg. Chem. 1999, 38, 6298.(31) Hamann, T. W.; Farha, O. K.; Hupp, J. T. J. Phys. Chem. C 2008,112, 19756.(32) Vijayalakshmi, R.; Rajendran, V. Arch. Appl. Sci. Res. 2012, 4,1183.(33) He, J.-A.; Mosurkal, R.; Samuelson, L. A.; Li, L.; Kumar, J.Langmuir 2003, 19, 2169.(34) Dell’Orto, E.; Raimondo, L.; Sassella, A.; Abbotto, A. J. Mater.Chem. 2012, 22, 11364.(35) Xin, X.; Scheiner, M.; Ye, M.; Lin, Z. Langmuir 2011, 27, 14594.(36) Peter, L. M.; Walker, A. B.; Boschloo, G.; Hagfeldt, A. J. Phys.Chem. B 2006, 110, 13694.(37) Williams, V. O.; Jeong, N. C.; Prasittichai, C.; Farha, O. K.;Pellin, M. J.; Hupp, J. T. ACS Nano 2012, 6, 6185.(38) Dunn, H. K.; Peter, L. M. J. Phys. Chem. C 2009, 113, 4726.

Figure 11. Energy level schematic of the ZT hybrid film showing theenergetics of the individual components used in the cell.

The Journal of Physical Chemistry C Article

dx.doi.org/10.1021/jp304622d | J. Phys. Chem. C 2012, 116, 23864−2387023870